Summary Low-salinity waterflooding (LSF) is one of the least-understood enhanced-oil-recovery (EOR)/improved-oil-recovery (IOR) methods, and proper understanding of the mechanism(s) leading to oil recovery in this process is needed. However, the intrinsic complexity of the process makes fundamental understanding of the underlying mechanism(s) and the interpretation of laboratory experiments difficult. Therefore, we use a model system for sandstone rock of reduced complexity that consists of clay minerals (Na-montmorillonite) deposited on a glass substrate and covered with crude-oil droplets and in which different effects can be separated to increase our fundamental understanding. We focus particularly on the kinetics of oil detachment when exposed to low-salinity (LS) brine. The system is equilibrated first under high-salinity (HS) brine and then exposed to brines of varying (lower) salinity while the shape of the oil droplets is continuously monitored at high resolution, allowing for a detailed analysis of the contact angle and the contact area as a function of time. It is observed that the contact angle and contact area of oil with the substrate reach a stable equilibrium at HS brine and show a clear response to the LS brine toward less-oil-wetting conditions and ultimately detachment from the clay substrate. This behavior is characterized by the motion of the three-phase (oil/water/solid) contact line that is initially pinned by clay particles at HS conditions, and pinning decreases upon exposure to LS brine. This leads to a decrease in contact area and contact angle that indicates wettability alteration toward a more-water-wet state. When the contact angle reaches a critical value at approximately 40 to 50°, oil starts to detach from the clay. During detachment, most of the oil is released, but in some cases a small amount of oil residue is left behind on the clay substrate. Our results for different salinity levels indicate that the kinetics of this wettability change correlates with a simple buoyancy- over adhesion-force balance and has a time constant of hours to days (i.e., it takes longer than commonly assumed). The unexpectedly long time constant, longer than expected by diffusion alone, is compatible with an electrokinetic ion-transport model (Nernst-Planck equation) in the thin water film between oil and clay. Alternatively, one could explain the observations only by more-specific [non- Derjaguin–Landau–Verwey–Overbeek (DLVO) type] interactions between oil and clay such as cation-bridging, direct chemical bonds, or acid/base effects that tend to pin the contact line. The findings provide new insights into the (sub) pore-scale mechanism of LSF, and one can use them as the basis for upscaling to, for example, pore-network scale and higher scales (e.g., core scale) to assess the impact of the slow kinetics on the time scale of an LSF response on macroscopic scales.